PURPOSE There is equipoise regarding the optimal treatment of clinical stage (CS) I nonseminomatous germ cell testicular cancer (NSGCT). Formal mechanisms that enable patients to consider cancer outcomes, treatment-related morbidity, and personal preferences are needed to facilitate decision making between retroperitoneal lymph node dissection (RPLND), primary chemotherapy, and surveillance. METHODS Decision analysis was performed using a Markov model that incorporated likelihoods of survival, treatment-related morbidity, and utilities for seven undesired post-treatment health states to estimate the quality-adjusted survival (QAS) for each treatment option. Utilities were obtained from 24 hypothetical NSGCT patients using a visual analog (rating) scale and standard gamble. Results Overall, QAS associated with each treatment was high and differences in QAS were small. Surveillance was the preferred intervention for patients with a risk of relapse less than 33% and 37% using the rating scale and standard-gamble method of utility assessment, respectively. Active treatment was favored over surveillance for patients with relapse risk on surveillance greater than 33% and 37% by the rating scale (RPLND preferred) and standard-gamble methods (primary chemotherapy preferred), respectively. Substantial differences in average utilities were seen depending on the method used. By the rating scale, patients substantially devalued life in six of seven undesired health states but they were surprisingly tolerant of treatment-related morbidity using standard gamble. CONCLUSION A decision model has been developed for CS I NSGCT that estimates QAS for RPLND, primary chemotherapy, and surveillance by considering cancer outcomes, morbidity, and patient preferences. Surveillance was the preferred intervention for all except those patients at high risk for relapse.